---
title: "DeepSeek V4 Flash 0731"
description: "DeepSeek V4 Flash 0731 by DeepSeek: 1.02M context. Compare provider prices and applicable pricing conditions on the model page. Pay as you go."
canonical: https://pirouter.ai/models/deepseek/deepseek-v4-flash
last-updated: 2026-10-15
---

# DeepSeek V4 Flash 0731

> DeepSeek V4 Flash 0731 by DeepSeek: 1.02M context. Compare provider prices and applicable pricing conditions on the model page. Pay as you go.

DeepSeek V4 Flash is the fast, low-cost model of the DeepSeek V4 generation, switching between deep thinking and plain chat. Text input with a 1M-token context, up to 384K output, tool calling, structured output and adjustable reasoning effort.

## At a glance

- Model ID: `deepseek/deepseek-v4-flash`
- Also accepted as: `deepseek-v4-flash`, `deepseek-v4-flash-0731`
- Vendor: DeepSeek
- Category: text
- Openness: Proprietary
- Model page: https://api-docs.deepseek.com/
- Context: 1,024,000 tokens
- Max output: 384,000 tokens
- Added to pirouter: 2026-07-31
- Status: Available

## Pricing (USD per 1M tokens)

|  | Input | Output | Cache read | Reasoning |
|---|---|---|---|---|
| Standard | $0.264 ~~$0.44~~ | $0.792 ~~$1.32~~ | $0.0084 ~~$0.014~~ | $0.792 ~~$1.32~~ |

40% off for a limited time · Ends 2026-10-31 23:59 HKT · Same price on every provider · this is what you are billed.

## Capabilities

- Function calling
- Tool choice
- Parallel tool calls
- Structured outputs
- Response schema
- Logprobs
- Reasoning
- Prompt caching
- System messages
- Streaming

## Reasoning effort

- max, xhigh, high, medium, low, minimal, none (default: medium)

## How to call it

One pirouter API key works with every API format below; create it in the console. Use one of these formats; formats not listed are not available for this model.

- API key: https://app.pirouter.ai/sign-in

### OpenAI Responses

- Endpoint: `POST /v1/responses` on https://api.pirouter.ai
- Auth: `Authorization: Bearer <key>`
- SDK base URL: https://api.pirouter.ai/v1
- Reference: https://pirouter.ai/docs/api/responses

```bash
curl https://api.pirouter.ai/v1/responses \
  -H "Authorization: Bearer $PIROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4-flash",
    "input": "Hello!"
  }'
```

### OpenAI Chat

- Endpoint: `POST /v1/chat/completions` on https://api.pirouter.ai
- Auth: `Authorization: Bearer <key>`
- SDK base URL: https://api.pirouter.ai/v1
- Reference: https://pirouter.ai/docs/api/chat-completions

```bash
curl https://api.pirouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $PIROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4-flash",
    "messages": [{ "role": "user", "content": "Hello!" }]
  }'
```

### Anthropic Messages

- Endpoint: `POST /v1/messages` on https://api.pirouter.ai
- Auth: `x-api-key: <key>`
- SDK base URL: https://api.pirouter.ai
- Reference: https://pirouter.ai/docs/api/messages

```bash
curl https://api.pirouter.ai/v1/messages \
  -H "x-api-key: $PIROUTER_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4-flash",
    "max_tokens": 1024,
    "messages": [{ "role": "user", "content": "Hello!" }]
  }'
```

## Providers

One row per provider; pirouter sends each request to a healthy one.

| Provider | API format | Context | Max output | Input /M | Output /M | Trains on your data | Data retention |
|---|---|---|---|---|---|---|---|
| volcengine | openai, anthropic | 1,024,000 | 384,000 | $0.264 ~~$0.44~~ | $0.792 ~~$1.32~~ | No | Not disclosed |

## Rankings

Where this model ranks on each blind-vote leaderboard, listing the boards where it makes the top 100. Open a board for the full ranking.

| Board | Rank | Score |
|---|---|---|
| [Overall](https://pirouter.ai/rankings/overall) | #84 / 385 | 1439 |
| [Coding](https://pirouter.ai/rankings/coding) | #89 / 380 | 1484 |
| [Math](https://pirouter.ai/rankings/math) | #79 / 371 | 1441 |
| [Creative writing](https://pirouter.ai/rankings/creative) | #75 / 383 | 1411 |
| [Instruction following](https://pirouter.ai/rankings/instruct) | #78 / 385 | 1434 |
| [Multi-turn](https://pirouter.ai/rankings/multiturn) | #71 / 383 | 1453 |
| [Long prompts](https://pirouter.ai/rankings/longctx) | #79 / 363 | 1449 |
| [Chinese](https://pirouter.ai/rankings/chinese) | #75 / 362 | 1480 |
| [Web dev](https://pirouter.ai/rankings/webdev) | #28 / 121 | 1581 |

Data: [LMArena leaderboard dataset](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset), used under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) · Updated 2026-10-01

## Supported parameters

`tools`, `tool_choice`, `parallel_tool_calls`, `response_format`, `structured_outputs`, `max_tokens`, `max_completion_tokens`, `temperature`, `top_p`, `seed`, `stop`, `logprobs`, `top_logprobs`, `reasoning`, `include_reasoning`, `stream`

## Links

- HTML: https://pirouter.ai/models/deepseek/deepseek-v4-flash
- JSON: https://pirouter.ai/api/models/deepseek/deepseek-v4-flash
- Providers JSON: https://api.pirouter.ai/v1/models/deepseek/deepseek-v4-flash/endpoints
- Leaderboard ranks (JSON): https://pirouter.ai/api/models/deepseek/deepseek-v4-flash/rankings
- Catalog: https://pirouter.ai/models
- Quickstart: https://pirouter.ai/docs/quickstart
- How routing works: https://pirouter.ai/docs/routing
- llms.txt: https://pirouter.ai/llms.txt

---

HTML version: https://pirouter.ai/models/deepseek/deepseek-v4-flash · Machine-readable index: https://pirouter.ai/llms.txt